Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Kan Twitter prediktera oljemarknadens framtida avkastningar?

Axelsson, Kristian LU (2013) NEKH01 20122
Department of Economics
Abstract
Is it possible to predict the future returns for the crude oil index by looking at the activity on Twitter? It is not possible to predict the market according to the efficient market hypothesis but earlier research on the topic states that it is possible for Twitter to predict the markets. This paper examines this phenomenon further by investigating the relationship between the market and Twitter activity during the “Arab spring”. The relationship is scrutinized by examining the number of times a particular keyword is mentioned on Twitter. The collected time series represents five keywords consisting of Egypt, Yemen, Syria, Kurdistan, and finally Pakistan. The relationship between the five time series and the future returns of crude oil... (More)
Is it possible to predict the future returns for the crude oil index by looking at the activity on Twitter? It is not possible to predict the market according to the efficient market hypothesis but earlier research on the topic states that it is possible for Twitter to predict the markets. This paper examines this phenomenon further by investigating the relationship between the market and Twitter activity during the “Arab spring”. The relationship is scrutinized by examining the number of times a particular keyword is mentioned on Twitter. The collected time series represents five keywords consisting of Egypt, Yemen, Syria, Kurdistan, and finally Pakistan. The relationship between the five time series and the future returns of crude oil index is examined using a Vector Autoregressive-model. The model is further tested by applying a Granger-non-causality test to examine the keywords predictive power. It turns out that the keywords have a predictive capability about four days prior to the change, which also can predict the direction of the WTI-index future returns. (Less)
Please use this url to cite or link to this publication:
author
Axelsson, Kristian LU
supervisor
organization
course
NEKH01 20122
year
type
M2 - Bachelor Degree
subject
keywords
WTI returns, Prediction, Arab spring, Twitter, Granger-causalirty
language
Swedish
id
3409274
date added to LUP
2013-02-12 15:41:06
date last changed
2013-02-12 15:41:06
@misc{3409274,
  abstract     = {{Is it possible to predict the future returns for the crude oil index by looking at the activity on Twitter? It is not possible to predict the market according to the efficient market hypothesis but earlier research on the topic states that it is possible for Twitter to predict the markets. This paper examines this phenomenon further by investigating the relationship between the market and Twitter activity during the “Arab spring”. The relationship is scrutinized by examining the number of times a particular keyword is mentioned on Twitter. The collected time series represents five keywords consisting of Egypt, Yemen, Syria, Kurdistan, and finally Pakistan. The relationship between the five time series and the future returns of crude oil index is examined using a Vector Autoregressive-model. The model is further tested by applying a Granger-non-causality test to examine the keywords predictive power. It turns out that the keywords have a predictive capability about four days prior to the change, which also can predict the direction of the WTI-index future returns.}},
  author       = {{Axelsson, Kristian}},
  language     = {{swe}},
  note         = {{Student Paper}},
  title        = {{Kan Twitter prediktera oljemarknadens framtida avkastningar?}},
  year         = {{2013}},
}